trclcomp {mvpart}R Documentation

Tree-Clustering Comparison

Description

This function compares the within-group variation for groups formed by tree partitioning and unconstrained clustering. The results are plotted and returned invisibly.

Usage

trclcomp(x, method = "com", km = TRUE, mrt = TRUE)

Arguments

x Rpart object with method "mrt" – see rpart
method The clustering method for the unconstrained clustering – see hclust
km If TRUE a K-Means clustering is compared with the multivariate tree partitioning.
mrt If TRUE an additional K-Means clustering with a starting configuration based on the multivariate tree partitioning is generated.

Details

The within-group variation for groups formed by multivariate tree partitioning and unconstrained clusterings are compared for all sizes of the hierarchy of tree partitions.

Value

Returns a list (invisibly) of the within-tree and within-cluster variation for all tree sizes.

Author(s)

Glenn De'ath

References

De'ath G. (2002) Multivariate Regression Trees : A New Technique for Constrained Classification Analysis. Ecology 83(4):1103-1117.

author{ Glenn De'ath }

Examples

data(spider)
fit <- mvpart(data.matrix(spider[,1:12])~herbs+reft+moss+sand+twigs+water,spider)  
trclcomp(fit)

[Package mvpart version 1.2-6 Index]